A machine learning-based approach for predicting renal function recovery in general ward patients with acute kidney injury

Aug 4, 2026

Upload by Aria

Acute kidney injury (AKI) is a major clinical challenge, yet renal function recovery—a key determinant of prognosis—has often been underemphasized. This study develops and validates a machine learning model to predict recovery of kidney function in general ward patients. Data from three hospitals include 350,345 cases, using Kidney Disease: Improving Global Outcomes criteria and recovery defined by a 33% serum creatinine decrease at AKI onset. The model demonstrates robust internal and external AUROC performance and uses SHAP values for interpretable predictors.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Abstract

Introduction

Methods

Study population

Outcome definition and features

Model development and validation

Interpretability analysis

Results

Conclusion

Original Article Kidney Res Clin Pract 2024;43(4):538-547 pISSN: 2211-9132 • eISSN: 2211-9140 [https://doi.org/10.23876/j.krcp.23.330](https://doi.org/10.23876/j.krcp.23.330) A machine learning-based approach for predicting renal function recovery in general ward patients with acute kidney injury Nam-Jun Cho1,*, Inyong Jeong2,*, Yeongmin Kim2, Dong Ok Kim2, Se-Jin Ahn2, Sang-Hee Kang3, Hyo-Wook Gil1 , Hwamin Lee2 1Department of Internal Medicine, Soonchunhyang University Cheonan Hospital, Cheonan, Republic of Korea 2Department of Medical Informatics, Korea University College of Medicine, Seoul, Republic of Korea 3Department of Surgery, Korea University Guro Hospital, Seoul, Republic of Korea Background: Acute kidney injury (AKI) is a significant challenge in healthcare. While there are considerable researches dedicated to AKI patients, a crucial factor in their renal function recovery, is often overlooked. Thus, our study aims to address this issue through the development of a machine learning model to predict restoration of kidney function in patients with AKI. Methods: Our study encompassed data from 350,345 cases, derived from three hospitals. AKI was classified in accordance with the Kidney Disease: Improving Global Outcomes. Criteria for recovery were established as either a 33% decrease in serum creatinine levels at AKI onset, which was initially employed for the diagnosis of AKI. We employed various machine learning models, selecting 43 pertinent features for analysis. Results: Our analysis contained 7,041 and 2,929 patients’ data from internal cohort and external cohort respectively. The Categorical Boosting Model demonstrated significant predictive accuracy, as evidenced by an internal area under the receiver operating characteristic (AUROC) of 0.7860, and an external AUROC score of 0.7316, thereby confirming its robustness in predictive performance. SHapley Additive exPlanations (SHAP) values were employed to explain key factors impacting recovery of renal function in AKI patients. Conclusion: This study presented a machine learning approach for predicting renal function recovery in patients with AKI. The model performance was assessed across distinct hospital settings, which revealed its efficacy. Although the model exhibited favorable outcomes, the necessity for further enhancements and the incorporation of more diverse datasets is imperative for its application in real-world. Keywords: Acute kidney injury, Hospital records, Machine learning, Recovery of function, Creatinine Introduction Despite treatment advancements in recent years, acute kidney injury (AKI) remains a significant concern in the medical field. AKI independently contributes to the escalation of healthcare costs and prolongation of hospitalization Received: December 4, 2023; Revised: February 27, 2024; Accepted: April 23, 2024 Correspondence: Hwamin Lee Department of Medical Informatics, Korea University College of Medicine, 73 Goryeodae-ro, Seongbuk-gu, Seoul 02841, Republic of Korea. E-mail: [hwamin@korea.ac.kr](hwamin@korea.ac.kr) ORCID: [https://orcid.org/0000-0002-6482-3511](https://orcid.org/0000-0002-6482-3511) *Nam-Jun Cho and Inyong Jeong contributed equally to this study as co-first authors. © 2024 by The Korean Society of Nephrology This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial and No Derivatives License ([https://](https://)[ ](https://)[creativecommons.org/licenses/by-nc-nd/4.0/](creativecommons.org/licenses/by-nc-nd/4.0/)) which permits unrestricted non-commercial use, distribution of the material without any modifications, and reproduction in any medium, provided the original works properly cited. periods, while also elevating the incidence of in-hospital complications and mortality rates [1–3] . This condition is recognized as one of the most prevalent diseases, exhibiting incidence rates of 10% to 15% in general hospital admissi

FAQs

How is renal function recovery defined in this study?

Recovery is defined as a 33% decrease in serum creatinine levels at the onset of acute kidney injury (AKI). This threshold was initially used for AKI diagnosis as well.

What data sources were used to build and test the model?

The internal cohort uses data from two hospitals, while the external cohort uses data from a separate hospital. Overall, the study includes 350,345 cases from three hospitals.

Which machine learning model performed best and how was it evaluated?

The Categorical Boosting Model showed significant predictive accuracy, with AUROC values of 0.7860 internally and 0.7316 externally. Performance was assessed across distinct hospital settings.

Aug 4, 2026

Upload by Aria